How it works

Private AI, running on a computer you own

SDL AI Studio installs on your machine and runs its models there. No server somewhere else doing the work by default, and no account required to start.

Local-first, plainly

Your machine does the work

Most AI tools send your prompts to a server you do not control. SDL AI Studio flips that around. The AI models run on your own computer, so your chats, projects, and documents stay there by default. There is no cloud step required just to use the app.

When you want more reach, you can switch on optional cloud models or web search for a specific chat. Those stay off until you choose them, so you always know what, if anything, leaves your machine.

Getting started

Three steps, start to finish

No server setup and no account to create before you can start working.

  1. Install SDL AI StudioThe installer sets up the local engine for you and downloads a starter model, so you have something to chat with right away.
  2. Pick your modelsChoose which local models to run based on your hardware. Smaller models are faster on modest machines, larger models want more RAM and, ideally, a GPU.
  3. Work privatelyChat, organize work into projects, research, and generate images, all on your machine. Turn on optional cloud tools only if and when you want to.
What it needs

Hardware expectations

Local AI runs on your hardware, so what you get out of it depends on what you put in.

Operating system
Windows, macOS, or Linux
Disk
Room for the models you choose, a few GB each
Processor and memory
A modern CPU works. More RAM helps when you run larger models.
Graphics, optional
A GPU is not required, but it makes local chat and image generation much faster

Local AI leans on your hardware more than a hosted tool does. A modest laptop will run smaller models comfortably. If you plan to use larger models or generate images often, more RAM and a GPU make a real difference.

SDL AI Studio checks your CPU, RAM, and GPU and gives you a simple performance rating, so you know what to expect. If replies feel slow, that is your hardware working hard, not the app misbehaving. Here is the rough picture of what different setups can handle.

ExcellentModern high-end GPU, for example an RTX 4070, 4080, or 4090. Fast chat and image generation.
GoodMainstream GPU, for example an RTX 3060 or 4060. Comfortable for everyday use.
BasicCPU only, no dedicated GPU. Chat works, image generation is slower.
LimitedOlder or low-power hardware. Best with smaller models and shorter tasks.